Gilbert + Tobin: 87% of Active Users on ChatGPT

The Australian firm Gilbert + Tobin reports 87% active users among those authorized to use ChatGPT. Management cites a case of an annual review for 120 partners completed in just a few minutes for about $62, with a governance framework that includes data residency in Australia and access controls. The stated goal is not workforce reduction, but rather freeing up time for high-touch activities.

A Quantified Case: An Annual Review for 120 Partners at $62

In a confidential annual review process involving 120 partners, ChatGPT Work was used to update files from a spreadsheet, generate PDFs, and assemble deliverables. According to management, the processing was completed in a matter of minutes. The cost of this operation is estimated at around $62, based on usage data from the tool. This amount is presented as a good exchange considering the time saved and the work that would have required several people over several days. Management emphasizes the importance of having appropriate administrative controls, with ChatGPT Work now providing detailed usage metrics.

Gilbert + Tobin Records 87% Active Users on ChatGPT

According to the firm, 87% of users with access to ChatGPT are actively using it, a rate that is more than double that observed for other tools. No usage quotas have been set; the priority has been given to alignment with individual missions. The business transformation team has presented workflows tailored to functions, complemented by videos, presentations, challenges, and an early AI bonus to generate ideas throughout the firm. Management believes that leaders must practice with these tools, experiment, and learn what works to train the organization.

Method and Safeguards: From Experimentation to Controlled Environments

From the moment ChatGPT was made available, teams were encouraged to experiment, with a strict rule prohibiting the entry of client data or confidential information. According to management, this constraint allowed for testing the tool in a safe environment. The firm then opened secure access to AI with ChatGPT Enterprise, first in a controlled environment and then expanded. Data residency in Australia, contractual clauses, role-based access control, and administrative tools provided the necessary confidence to scale while meeting internal and client requirements. The guiding principle is clear: AI is a tool to support judgment; everyone must define the task, verify the output, and take responsibility for the deliverable. The progression is described in three phases: experimentation, personal productivity, and then organizational productivity.

Concrete Use Cases: A Managerial "Twin" and Plugins

The CEO has designed a "digital twin" through a customized GPT, now transitioning to Plugins, based on his writings, feedback received, internal information, and his working style. Initially personal, the tool became more useful once shared with his direct collaborators, who use it to receive feedback aligned with his style via ChatGPT. It serves to test thoughts in the right context before a discussion, without making decisions or speaking on his behalf. At the firm level, the introduction of ChatGPT Enterprise, ChatGPT Work, and Codex aims to shift from individual assistance to multi-step agentic workflows, accompanied by human review and approval.

Stated Objectives and Desired Value

The stated objectives are to improve client services, make the firm an excellent workplace, and establish itself as a market leader. Management indicates that it is not primarily seeking cost savings or workforce reduction, but rather freeing up time for activities where relationships and guidance matter. This initiative takes place within a major corporate law firm in Australia, active in capital markets, mergers and acquisitions, litigation, and technology, within a profession that relies on expertise, judgment, and the protection of sensitive information.

Proposed Managerial Roadmap for Leaders

Management recommends that CEOs adopt these tools themselves, experiment with real tasks, understand their limitations, and value learning. It advocates for pairing this exemplarity with clear governance and practical empowerment: role-appropriate examples, safe spaces for testing, and maintained responsibility for outcomes. According to them, the stakes go beyond task acceleration and can improve the quality and volume of outputs while freeing up time for complex and strategic work.